Research On Resource Allocation Of NFV Optical/electronic Hybrid Network

In recent years,with the development of 5G,Cloud Computing,Internet of Things(Io T),the traditional network devices,which is tightly coupled with hardware and software,has inherent deficiencies in scalability and flexibility,so it is unable to meet the complex and diverse services in the new era.On the one hand,Network Function Virtualization(NFV)allows network functions to run on the general computing devices in the form of software,which not only reduces the operation and maintenance costs,but also provides services flexibly and quickly.However,the bandwidth requirements of services change greatly and the electronic networks cannot provide enough bandwidth.Although the optical networks can do it,it has insufficient flexibility.Therefore,this thesis introduces an optical/electronic hybrid network,which combines the advantages of large capacity and low power consumption of optical networks,and also has the characteristics of electronic networks.In the NFV networks,the service is the Service Function Chain(SFC).The network needs to allocate computing resources and bandwidth resources for Virtual Network Function(VNF)and virtual link of SFC.Most of the studies do not take the bandwidth of the underlying network into consideration,which makes these studies cannot be directly applied to the optical/electronic hybrid network scenario.Hence,this thesis mainly studies how to do the optimal SFC mapping and resource allocation in optical/electronic hybrid network.In this thesis,the mathematical model of SFC mapping problem of hybrid optical/electronic network is established with the goal of utility.The utility function is mainly obtained by subtracting the cost of servers and links from the utility that getting from the VNFs on node.Because the VNF mapping problem is NP hard,this paper uses Markov approximation to obtain the near optimal solution of the original problem.The algorithm is provided to adaptively embed SFC requests with the maximal utility function.Simulation results demonstrate the quality of the proposed algorithms in terms of utility function and the adaptivity of the proposed algorithm.Finally,based on the dynamic multiple requests scenario,the energy consumption model of NFV optical/electronic hybrid network is established.Then an Integer Linear Programming(ILP)model and three heuristic algorithms are introduced.These three heuristics all achieve energy saving by shutting down idle devices during resource allocation,and trade-off between energy consumption and SFC requests acceptance rate.After comparing with ILP by simulation,results show that the three heuristics are effective.